标题：Nonparametric and Doubly Robust Inference for Continuous Treatment Effects

摘要:

In this talk, I will review recent advances in causal inference with continuous treatments, focusing on the estimation of the average dose-response function (ADRF).

Under the potential outcomes framework, existing works develop nonparametric methods, including local linear kernel regression (LLKR) and empirical risk minimization (ERM), to achieve doubly robust estimation based on augmented inverse probability weighting (AIPW) scores.

These estimators enable flexible modeling without restrictive parametric assumptions, while retaining desirable theoretical properties under mild conditions.